IZA DP No. 322

Works Councils and Collective Bargaining inGermany: The Impact on Productivity and Wages

Olaf HüblerUwe Jirjahn

DI

SC

US

SI

ON

PA

PE

R S

ER

IE

S

Forschungsinstitutzur Zukunft der ArbeitInstitute for the Studyof Labor

July 2001

!

"#$%

Discussion Paper No. 322 July 2001

IZA

P.O. Box 7240 D-53072 Bonn

Germany

Tel.: +49-228-3894-0 Fax: +49-228-3894-210

Email: iza@iza.org

This Discussion Paper is issued within the framework of IZA’s research area Anyopinions expressed here are those of the author(s) and not those of the institute. Research disseminated by IZA may include views on policy, but the institute itself takes no institutional policy positions. The Institute for the Study of Labor (IZA) in Bonn is a local and virtual international research center and a place of communication between science, politics and business. IZA is an independent, nonprofit limited liability company (Gesellschaft mit beschränkter Haftung) supported by the Deutsche Post AG. The center is associated with the University of Bonn and offers a stimulating research environment through its research networks, research support, and visitors and doctoral programs. IZA engages in (i) original and internationally competitive research in all fields of labor economics, (ii) development of policy concepts, and (iii) dissemination of research results and concepts to the interested public. The current research program deals with (1) mobility and flexibility of labor markets, (2) internationalization of labor markets and European integration, (3) the welfare state and labor markets, (4) labor markets in transition, (5) the future of work, (6) project evaluation and (7) general labor economics. IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion. Citation of such a paper should account for its provisional character.

IZA Discussion Paper No. 322 July 2001

&'(&

∗

This paper investigates the interaction between establishment-level codetermination and industry-level collective bargaining in Germany. Based on a simple bargaining model we derive our main hypothesis: In establishments covered by collective bargaining agreements works councils are more likely to be engaged in productivity enhancing activities and less engaged in rent seeking activities than their counterparts in uncovered firms. Using data from German manufacturing establishments, our empirical analysis confirms this hypothesis. The presence of works councils exerts a positive impact on productivity within the covered industrial relations regime but not within the uncovered industrial relations regime. In contrast, the presence of works councils has a positive effect on wages within the uncovered industrial relations regime but not to the same degree within the covered industrial relations regime. JEL Codes: D23; J24; J31; J51; J53 Keywords: Dual industrial relations; bargaining; efficiency; rent seeking; correlated

double selection Olaf Hübler Institute of Quantitative Economic Research University of Hannover Königsworther Platz 1 30167 Hannover Germany Tel.: +49 511 762 4794 Fax: +49 511 762 3923 Email: huebler@mbox.iqw.uni-hannover.de

∗ We are grateful to Kornelius Kraft, Friedhelm Pfeiffer, Claus Schnabel, Stephen C. Smith, Joachim Wagner and the participants of the 9th colloquium of the DFG long-term research program “Industrial Economics and Input Markets” for helpful comments. .

1

The German system of industrial relations is characterized by a dual structure of

employee representation through works councils and unions. Works councils provide

a highly developed mechanism for establishment-level participation while collective

bargaining agreements are negotiated between unions and employers’ associations on

an industrial level. The basic question addressed in this paper is: Does the impact of

works councils on productivity and wages differ between establishments covered by

collective bargaining agreements and establishments not covered by collective

bargaining agreements?

Empirical evidence and a theoretical analysis how codetermination and collective

bargaining interact in the process of the creation and distribution of economic rents,

are extremely scarce. Almost all of the somewhat sparse econometric studies examine

the isolated impact of works councils on firm performance. This little knowledge is

unsatisfying for at least two reasons: First, German councils have attracted

considerable attention in the United States as an alternative form of worker

participation to promote industrial democracy and economic performance. Also in

Germany there is a recent discussion on the role of works councils because the

German government plans to strengthen their participatory rights. However, the

economic effects of works councils depend on other parameters of the industrial

relations system such as collective bargaining. Policies to encourage councils should

take into account these factors. Second, there is an ongoing discussion on the

economic effects of centralised bargaining. One claim is that centralised bargaining is

not sufficiently responsive to local conditions (Freeman and Gibbons 1995, Lindbeck

and Snower 1997). However, the interaction between codetermination and collective

2

bargaining, considered in this paper, sheds light on an indirect effect of centralised

bargaining that may contribute to increased firm performance. Even if productivity-

enhancing work practices must be negotiated on the establishment level between

management and works councils, these changes are more easily negotiated, when

substantial distributional conflicts are moderated on a central level by unions and

employers’ associations. The literature to date has not well recognised this aspect.

In a model by Freeman and Lazear (1995) the works council is a mechanism for

building trustful industrial relations and for improving the information flow between

workers and management. Without explicitly analysing the effects within a bargaining

framework, Freeman and Lazear also recognise that codetermination may strengthen

workers’ bargaining power. FitzRoy and Kraft (2000) extend this model to the system

of board-level co-determination.

In our paper the two faces of works councils are incorporated in a bargaining

model. On the one hand, works councils foster trust and cooperation. They provide a

mechanism for negotiating productivity-enhancing work practices. On the other hand,

councils can use their codetermination rights to hinder decisions if management and

the workforce fail to reach an agreement. Codetermination weakens the

management’s disagreement position. The presence of councils may result in both the

creation and the redistribution of economic rents. We argue that the management’s

disagreement position is less weakened in establishments covered by collective

bargaining agreements. Comparing covered and uncovered firms, the presence of

councils should have a more substantial impact on productivity and a less intense

impact on wages in the covered industrial relations regime.

Our empirical analysis is based on German establishment data from the ‘Hannover

Panel’. Applying a double-selection approach, we take into account the endogeneity

3

of the two central variables: The presence of a works council and the coverage by a

collective bargaining agreement. The results confirm our hypothesis. The presence of

works councils exerts a positive impact on productivity within the industrial relations

regime covered by collective bargaining agreements but not within the uncovered

industrial relations regime. In contrast, the presence of works councils exerts a

positive impact on wages within the uncovered industrial relations regime but less

within the covered industrial relations regime. These results complement our studies

about the determinants of pay-for-performance schemes, which obtain a positive

impact of councils on the use of performance pay and profit sharing in the covered

regime but not in the uncovered regime (Heywood, Hübler and Jirjahn 1998;

Heywood and Jirjahn 1999).

A related question has been investigated by FitzRoy and Kraft (1990). In their

study the interaction of works councils with union density exerts a negative impact on

the firm’s innovative activities. However, the focus of our approach is not on the

interaction of works councils with the proportion of the firm’s employees who are

union members. Rather, the basic hypothesis of this paper is that the impact of works

councils on firm performance depends on the moderation of substantial distributional

conflicts by unions and employers’ associations. Therefore, we investigate the

interaction of works councils with the coverage by collective bargaining agreements.

Our results also provide a counterpoint to empirical studies with Anglo-Saxon

data, which examine the interaction effects between decentralised collective

bargaining and methods of employee involvement such as consultative committees or

teams. For the United States Cooke (1994) finds that teams contribute substantially

more to value added in unionised firms than in non-union firms. Although comparing

the effects of worker participation in Britain and Germany, Addison, Siebert, Wagner

4

and Wei (2000) investigate interaction effects only with the British data. They obtain

positive effects of employee involvement on productivity only in non-union plants.

The plan of the article is as follows. In section II we present a non-technical

discussion of the dual industrial relations system. Section III develops a formal model.

Section IV describes the data source and the econometric methods. Section V presents

the results. Section VI concludes.

!!"""# $ %

The rights of the works councils are laid down in the Works Constitution Act (WCA).

The WCA was introduced in 1952 and reformed in 1972 and 1989. Councils shall be

elected by the whole work force of establishments with five or more permanent

employees. However, their creation depends on the initiative of the establishment’s

employees. Hence, councils are not present in all eligible firms. Council presence

depends on factors such as firm size, firm age and the gender composition of the work

force (Addison, Schnabel and Wagner 1997). The powers of the works councils can

be quite broad. They have full codetermination rights on a set of issues, including the

introduction of new payment methods, the fixing of job and bonus rates, the allocation

of working hours, the introduction and use of technical devices designed to monitor

employee performance, and up- and down grading. In these areas management cannot

take action without the agreement of the works council. The councils have less strong

consultation rights in matters such as changes in equipment and working methods that

affect job requirements. Their participation rights in financial and economic matters

cover information provision.

5

The WCA ties work councils to the general obligation to co-operate with

management ‘in a spirit of mutual trust . . . for the good of the employees and of the

establishment’. This calls for an economic explanation: How can codetermination

enhance joint firm surplus? First, in Germany works councils may exert a collective

voice role (FitzRoy and Kraft 1987; Frick and Sadowski 1995). Many working

conditions are workplace public goods (Freeman and Medoff 1984). Works councils

communicate worker preferences to management helping to optimise the provision of

those public goods.Second, workers will withhold effort and cooperation when an

employer cannot credible commit to take into account the workers’ interests. For

example, workers fearing job loss due to technological change do not reveal

information about potential productivity-enhancing innovations. Another well known

example is the ratchet effect. Workers, receiving pay for performance, will withhold

effort when they anticipate that the employer will alter the future terms of the

payment scheme in light of the workers’ past performance. Providing works councils

with codetermination rights is one mechanism for protecting the interests of the work

force (Smith 1991; Freeman and Lazear 1995). Hence, codetermination may foster the

workers’ cooperation with the introduction of efficiency-enhancing work practices,

including pay for performance and training for employees.

There may be other solutions to the commitment problem. Under some

circumstances reputation concerns induce an employer to act honestly (Bull 1987).

However, the reputation mechanism fails if an employer overly discounts the future

loss of trust and co-operation.1 In this situation a councils with veto rights can protect

the workers’ interests. Moreover, the reputation mechanism fails if workers have not

enough information to verify whether an employer behaves honestly or not (Kreps

1990). This suggests that the comprehensive information rights of the works council

6

and the reputation mechanism may be complementary. Finally, even if the owners of

the firm care about their reputation, the objectives of the managers and superiors do

not necessarily coincide with the owners’ interest in building trustful and cooperative

relationships with the employees. Recent literature has stressed the role of rent-

seeking activities of supervisors (Prendergast 1999). Councils may be an alternative

mechanism for restricting superiors’ favouritism and influence activities. They foster

the introduction of general rules for performance evaluation and promotion.

In sum, there are reasonable arguments that the presence of councils can enhance

joint firm surplus. This raises the question why not all firms take the initiative to

establish works councils. There are serveal answers to this question.2 First,

establishing work councils involves costs. The employees have to spend time and

effort to introduce a council. The employer has to bear expenses such as the release

time of works councillors. Second, councils may not only have an impact on the size

but also on the distribution of firm surplus. Co-determination may strengthen the

bargaining power of the workforce. The owners of a firm will informally resist to

establish councils if the redistribution of surplus dominates. Moreover, there will be

resistance by rent-seeking supervisors if they loose power due to co-determination.3

The aim of the WCA clearly is to restrict distributional conflicts on the

establishment level. Works councils are subject to a general peace obligation. They do

not have the right to strike. If council and management fail to reach an agreement,

they may appeal to an internal arbitration board or to the labour court. Moreover,

wage negotiations between councils and management are not authorised by law.

However, formal or informal establishment-level negotiations play an important role.

First, the introduction of potential productivity-enhancing work practices must be

negotiated when there are conflicting interests over how these practices are

7

established. For example, the reorganisation of work may require additional effort of

the workforce to improve productivity. In this case workers want to be compensated

for the increased disutility from working. Additionally, they want to share the

increased rents of the firm. Even if codetermination rights would not increase the

workers’ bargaining power, a positive impact of works councils on wages or fringe

benefits due to the increased surplus can be expected.

Second, codetermination may enhance the workers’ share in a given surplus.

Councils can use their codetermination rights on social or personnel matters to obtain

employer concessions on issues where they have no legal powers. As Müller-Jentsch

(1995, p. 60) puts it, ‘For example, the council can hold up decisions on staff

movements where it has veto rights, or it can withhold consent on overtime where it

has co-determination rights, in order to obtain concessions on other issues’.

Codetermination weakens the management’s disagreement position. If management

and works council fail to reach an agreement in the wage negotiations, the council can

threaten to be uncooperative in areas where its consent is necessary. Even if these

negotiations are informal, the outcome can be binding for the negotiating parties. The

council may commit to co-operate with management by signing work agreements on

issues where it has legal rights. Management may commit to pay higher wages by

placing workers in higher wage groups.4 Indeed, most of the somewhat sparse

econometric studies find a positive impact of councils on the wage level (Jirjahn and

Klodt 1999) and a negative effect on the wage differential between skilled and

unskilled workers within establishments (Hübler and Meyer 2000). The last result

suggests that unskilled workers gain disproportionately through their assignment to

higher wage groups. In sum, there are reasonable arguments that codetermination

results in a redistribution of firm surplus. Moreover, the workers’ increased

8

bargaining power makes the impact of councils on work practices ambiguous. On the

one hand, councils foster the trust that is necessary for establishing productivity-

enhancing work practices. On the other hand, they may use their bargaining power to

negotiate less productive work practices that require less effort.

Codetermination will more likely result in higher productivity and in less intense

wage increases when the councils are less engaged in rent-seeking behaviour. One

crucial factor, that has an impact on the negotiations between works councils and

management, is the coverage by a collective bargaining agreement. Collective

agreements are usually negotiated between unions and employers’ associations on an

industrial level. Hence, firms are typically covered that are members of an employers’

association. Freeman and Lazear (1995, pp. 31-32) argue that councils and

management maximise joint surplus when there exists an exogenously fixed sharing

scheme that divides the surplus. Codetermination fits better to an industrial relations

system where distributional conflicts are moderated outside the firm. However,

collective bargaining does not provide a surplus-sharing scheme for dividing firm

surplus. It fixes a certain wage level and general working conditions.

In this paper we offer an alternative approach: The employers’ associations support

management with expertise in case that there are questions of law or lawsuits.

Therefore, the opportunities for a council to obtain employer concessions on wages by

withholding co-operation in areas where it has codetermination rights are more

restricted in covered establishments. Moreover, not only employers’ associations but

also unions are interested in preventing councils from rent seeking activities. First,

establishment-level bargaining undermines the unions’ power and status and

contributes to dispersed earnings across firms. Second, the unions’ interests transcend

those of the workforce in an individual firm. Because of the centralised system of

9

collective bargaining unions are interested in the industry- or nation-wide

employment level. Svejnar (1982) goes so far to assume that the unions’ influence

changes the objective function of the works councils. However, even if the councils

only represent the interests of the insiders within the establishments, unions can

impose restrictions on the councils’ activities. A union may provide a council with

expertise to strengthen the council’s position against an opportunistic employer but it

will prevent rent seeking activities. In sum, we expect that the management’s

disagreement position is less weakened due to codetermination when the

establishment is covered by a collective bargaining agreement. Therefore, works

councils are more likely to have a positive impact on productivity and a less strong

effect on wages in establishments covered by collective agreements.

&"# $ %

Our model illustrates three points: First, council presence and coverage by a collective

agreement are endogenous. Second, councils have an impact on the creation and

distribution of rents. Third, the outcome of the establishment-level negotiations

depends on the coverage by a collective agreement.

' &

We consider a firm with N identical workers. Employer and workers play a three-

stage game. At the first stage the employer decides on the membership in an

employers’ association. Membership in an employers’ association imposes costs CB

on the employer. The employer has to pay a membership fee. Moreover, the coverage

10

by a collective bargaining agreement may impose inflexible restrictions. At the

second stage workers decide on establishing a works council after observing the

employer’s decision. The creation of a council imposes costs CW on the work force.5

The election of a council requires co-ordination and communication among the

workers. Additionally, workers have to overcome the informal resistance of rent-

seeking supervisors. Finally, at the third stage, bargaining may take place and output

is produced. The general form of the production function is given by

(1) Q(e, α, N)= (1+ e)F(N) if there is an agreement,

F(N) if there is a conflict.α

where F(.) denotes the basic technology, e is our term for work practices and α,

0 ≤ ≤α 1, is related to the council’s opportunities to hinder decisions. With this

production function we capture the two faces of works councils and combine two

strands of the literature.

Following McCain (1980), we assume that bargaining over both work practices

and wages is only possible when a works council is present.6 Without a council

workers will not co-operate with the introduction of new work practices because they

fear employer opportunism. Therefore, e is equal to zero. However, the presence of a

works council fosters trust and co-operation. The introduction of new work practices

can be negotiated. In this case e may be nonzero. If e is positive, this term represents

productivity-enhancing work practices. We also allow for the case that e is negative.

In this case the firm introduces work practices that result in lower productivity.

Employees may prefer work practices that improve occupational health or reduce

environmental pollution affecting their families (Askildsen, Jirjahn and Smith2000).

11

In models of wage drift α captures work-to-rule actions in case that employer and

union fail to reach an agreement (Moene, Wallerstein and Hoel 1993, pp. 100-103). In

our model this term is related to the council’s opportunities to hinder decisions if

management and works council cannot reach an agreement. A small α represents a

situation where a council has a strong power to disrupt production. In particular we

assume

(2) α =

1 if no council is present,

if a council is present in an uncovered establishment

with 0 < < 1 if a council is present in an covered establishment.

0

~ ~

,

α α

Codetermination increases the workers’ bargaining power by weakening the

employer’s position in case of a disagreement.7 For convenience we normalise α to be

equal to zero if there is a council in an establishment not covered by a collective

bargaining agreement. The arguments discussed in section II suggest that the

opportunities to hinder decisions are more limited if the establishment is covered by a

collective bargaining agreement. This is captured by 0 < ~α < 1.

At the third stage, each worker’s utility function has the Stone-Geary form

(3) U(e, w)= ( e)wθ −

where w denotes the wage. Workers clearly like wages. A negative value of e

increases utility due to the improvement in occupational health. A positive value of e

decreases utility because productivity enhancing work practices require more effort.

The parameter θ is the reference level for productivity-enhancing work practices

above which the worker does not wish to work. On the one hand, it depends on the

worker’s personal characteristics. On the other hand, also general firm characteristics

12

are important.8 In this respect we may say that favourable conditions for investing in

work practices are represented by a large θ.

In case of a disagreement the firm employs the workers by paying them their

reservation utility. Each worker’s reservation utility is normalised to be equal to zero.

Consequently, the firm’s profit at the second stage is

(4) π(e, α, N, w)= (1+ e)F(N) wN if there is an agreement

F(N) if there is a conflict

−

α

( %

In case that no council is present we have α = 1 and e = 0. If the firm does not invest

in work practices it can employ the workers by paying them a wage equal to zero. Of

course, workers are interested in higher wages. However, the employer does not

benefit from paying higher wages. Workers neither can hinder decisions nor can

achieve a higher utility by leaving the firm. Moreover, new work practices cannot be

negotiated due to the lack of trust and co-operation. In sum, this situation is no

bargaining situation. The outcome is the same for a covered firm and an uncovered

firm. We have w* = 0 and e* = 0.

In contrast, if a works council is present we have a bargaining situation.9 First, the

council can threaten to hinder decisions in case of a conflict. Second, works council

and employer can negotiate both wages and work practices. The Nash product

(5) Ω = [( e)wN] [(1+ e)F(N) wN F(N)]θ α− ⋅ − −

is maximised by choosing w and e. We obtain

13

(6a) w = F(N)

3N[ + (1 )]θ α−

(6b) e = 1

3[2 (1 )]θ α− −

The term e for work practices is increasing in θ. Obviously, the parties invest more in

the improvement of work practices if the conditions are favourable. Moreover, the

wage rate w is increasing in θ. The council has an impact on wages due to the

increased firm surplus. Consider the case α = 1. Even if codetermination would have

no impact on the workers’ bargaining power we would observe higher wages in

establishments with a works council because productivity enhancing work practices

are introduced. However, codetermination increases the workers’ bargaining power.

The term (1 )− α captures this effect. The wage rate w is increasing in (1 )− α .

Codetermination results in higher wages by strengthening workers’ bargaining power.

Moreover, the impact of council presence on work practices is ambiguous. If

0.5 (1 )− α > θ we obtain e < 0. If 0.5 (1 )− α > θ we obtain e > 0. In sum, e is

decreasing in (1 )− α and w is increasing in (1 )− α . We will observe a more

substantial impact of councils on productivity and a less intense impact on wages if

the councils’ opportunities to hinder decisions are restricted.

Let us consider the case that the firm is not covered by a collective bargaining

agreement. Inserting α = 0 into (6a) and (6b) yields

(7a) w** = F(N)

3N[ +1]θ

(7b) e** = 1

3[2 ]θ −1

In case that the establishment is covered we have α α= ~ with 0 < ~α < 1. This yields

14

(8a) w*** = F(N)

3N[ + (1 )]θ α− ~

(8b) e*** = 1

3[2 (1 )]θ α− − ~

Comparing (7) with (8), the presence of a works council results in an increase of the

wage rate that is more pronounced if the firm is not covered by a collective

agreement. A positive impact of work councils on productivity is more likely in a

covered firm. In case of an uncovered firm codetermination results in productivity-

enhancing work practices if θ > 0.5. In case of a covered firm co-determination results

in productivity-enhancing work practices if θ > 0.5(1- ~α ).

) "

At the first stage the employer’s decision to become a member of an employers’

association depends on CB and on the impact of the coverage on the establishment-

level negotiations. After observing the employer’s decision the creation of a works

council by the workers depends on the costs CW and on the work practices and wages

negotiated at the second stage. Since the works council’s opportunities to hinder

decisions are more restricted in an establishment covered by a collective agreement

the work force can obtain a larger utility be creating a council if the employer is no

member of an employers’ association. The employer anticipates that the council’s

bargaining power is reduced in a covered establishment. The structure of the game

and the payoffs are shown in Figure 1. It depends on CB, CW, F(N), θ and ~α which

strategy pair, (COLLECT = 0, WOCO = 0), (COLLECT = 1, WOCO = 0),

(COLLECT = 0, WOCO = 1) or (COLLECT = 1, WOCO = 1), is the subgame-perfect

15

equilibrium. The conditions are shown in Table 1. The Figures 2.a-c illustrate the

equilibria for various parameter constellations.

The interpretation of the conditions for this case is

straightforward. Even if the establishment is not covered by a collective agreement

the workers’ gains from establishing a works council are smaller than the costs CW.

The costs of creating a council are not offset by the increase in utility due to a higher

wage rate. Since the employer anticipates that the workers will not create a works

council there is no reason to become a member of an employers’ association.

This strategy pair is a subgame-perfect equilibrium

if it pays for the employer to discourage the workers from creating a council. The

workers would only create a council in case that the establishment is not covered by a

collective agreement. By bearing the costs CB of a coverage the employer avoids

higher wages and possibly less productive work practices because it does not pay for

the workers to create a council in a covered establishment. Note that this equillibrium

is only possible for a small θ (θ < 2).

This strategy pair is a subgame-perfect equilibrium

if the employer cannot discourage workers from creating a works council. However, it

pays for the employer to reduce the council’s bargaining power by becoming the

member in an employer’s association.

This strategy pair is a subgame-perfect equilibrium

in two different situations. In the first situation, workers will create a council even if

the establishment is covered by a collective agreement. For the employer it does not

pay to reduce the council’s bargaining power because the costs CB are too large. In the

second situation, workers will elect a council only if the establishment is not covered

16

by a collective agreement. Again, the costs CB discourage the employer from the

membership in an employers’ association. Hence the workers create a works council.

The interpretation of these results is straightforward. Workers will only elect a

works council if the costs of establishing a council are offset by the increase in utility

due to a higher wage rate. On the one hand, the wage rate increases because workers’

bargaining power is strengthened. On the other hand, the wage rate increases when

the rent that can be distributed increases. Large firm size N as well as large output

F(N) and favourable conditions θ for productivity-enhancing work practices make a

council more likely whereas high costs of election CW make them less likely. The

employer will be the member of an employers’ association if the costs of the coverage

by a collective agreement are offset by the advantage of discouraging the creation of a

council or restricting rent seeking behaviour of a works council. Hence establishments

are more likely to be covered by a collective agreement if ~α F(N) is large and the

costs CB imposed by the collective agreement are low. In sum, council presence and

the coverage by a collective bargaining agreement are endogenous and depend on

establishment characteristics, worker characteristics and the institutional framework

of the industrial relations system.

As shown in Table 1 each strategy pair can be characterised by the corresponding

wage rate, productivity and profitability. To make the predictions of the model clear

let us consider four establishments that only differ in CB and CW. The subgame-

perfect equilibrium for the first firm is (COLLECT = 0, WOCO = 0), for the second

(COLLECT = 0, WOCO = 1), for the third (COLLECT = 1, WOCO = 0) and for the

fourth (COLLECT = 1, WOCO = 1). We can now derive our hypotheses about the

effects of works councils in covered and uncovered establishments. The presence of a

works council increases the wage. The impact is more pronounced in the

17

establishment not covered by a collective bargaining agreement. Depending on θ the

presence of a works council may result in a decreased or increased productivity.

Similarly, the presence of a council may result in smaller or larger profitability.

However, a positive impact on productivity and profitability is more likely in the

establishment covered by a collective bargaining agreement.

We stress that the focus of our model is on the indirect effect of collective

bargaining. Coverage by a collective bargaining agreement has no direct effect but an

indirect impact on wages by restricting rent seeking activities of works councils.

Fitzenberger and Franz (1999) note that the outcome of collective negotiations is

usually extended to the overwhelming number of employees in an industry. So we

may expect that centralised collective bargaining has an impact on wages at the

industry level while there are not necessarily wage differentials between covered and

uncovered establishments within an industry. Indeed, empirical studies with

establishment data usually do not find a significant impact of the coverage by

collective agreements on establishment wages (Jirjahn and Klodt 1999).

* # !&

Our empirical investigation is based on the Hannover Panel (Brand, Carstensen,

Gerlach and Klodt 1996), a four wave panel with data from manufacturing

establishments in Lower Saxony with 5 or more employees. The first questionnaire

was completed in autumn 1994. The number of establishments taking part in the study

declined from 1,025 (1994) to 849 (1995), 721 (1996) and 709 (1997). The sample is

stratified according to firm size and industry. The data were collected in face to face

interviews with firms owners or top managers. The questionnaire covered various

18

aspects of firm structure, firm behaviour and firm performance with an emphasis on

issues relating to personnel.

The objective of our empirical study is to determine the effects of works council

presence (WOCO) and the coverage by a collective agreement (COLLECT) on

productivity and wages. As wage variable (WAGE) we use wages and salaries per

employee and year in an establishment. Productivity (PROD) is measured by sales per

year minus the cost of raw material, consumables and supplies and of purchased

merchandise divided by the number of employees. Information on profits is not

available from the Hannover Panel. We use the employer’s quasi rent (QRENT)

determined by (sales – raw material etc. – wages)/number of employees as a proxy for

profit. The information on our central variable WOCO is only available from the first

and the third wave for the years 1994 and 1996. Therefore, the analysis is restricted to

those years. Information on WAGE, PROD and QRENT are available for 1994 and

1996 from the second and the fourth wave. The analysis is based on the pooled data

for both years.

Addison, Schnabel and Wagner (1998) and Addison, Siebert, Wagner and Wei

(2000) have also used parts of this data. Our paper differs in several respects from

those study about works councils. Using only the first wave of the Hannover Panel,

those studies examine the isolated impact of works councils on firm performance. In

contrast, this paper analyses the interaction of works councils with collective

bargaining for both years. Moreover, in our analysis we take into account the

endogeneity of our central variables WOCO and COLLECT. Finally, our

specifications are model-based.

We start with OLS estimates of wage equations where WOCO=:d1 and

COLLECT=:d2 are included as dummies

19

(9) y1 = WAGE = x1′β1 + d1 γ1 + d2 γ2 + u1.

Analogously, the productivity function (PROD) can be estimated

(10) y2 = PROD = x2′β2 + d1γ1* + d2γ2* + u2.

In the next step we test whether the WOCO on the one hand and COLLECT on the

other hand are correlated with the disturbance term (u). If unobserved determinants of

WOCO and COLLECT (v(WOCO) =: v1, v(COLLECT) =: v2), respectively, are

partially the same as in our objective function (9) and (10), respectively, the estimated

effects of the two variables in which we are interested are inconsistent. It is assumed

that the densities h1(v1, u1) and h2(v2, u1), respectively, are bivariate normal. Then

E(u1|WOCO, COLLECT, x) is nonzero but can be determined by probit estimates of

WOCO and COLLECT. Based on the following latent models for the WOCO and

COLLECT decision

(11) I1 = z1′α1 - v1

(12) I2 = z2′α 2 - v2

the dummies WOCO and COLLECT are defined by

(13) d1 = WOCO = 1 if I

otherwise

1 >

0

0

(14) d2 = COLLECT = 1 if I

otherwise

2 >

0

0

20

We have to distinguish between two cases. Either we assume that v1 and v2 are

independent or a correlation is allowed. In the first case as the conditional mean of u1

follows

(15) E(u1 | d1 , d2 , x1) =

σv1u1[d1(-ϕ(z1′α1/σv1) /Φ(z1′α1/σv1)) + (1-d1) (ϕ(z1′α1/ σv1 )/(1-Φ(z1′α1/σv1)] +

σv2u1[d2(-ϕ(z2′α2/σv2) /Φ(z2′α2/σv2)) + (1-d2) (ϕ(z2′α2/ σv2 )/(1- Φ(z2′α2/σv2)] ≡

δ1λ1 + δ2λ2.

An extension of equation (11) and (12) relaxes the restriction that v1 and v2 in the

index functions for WOCO and COLLECT decisions are independent. This seems

empirically plausible if some unobserved determinants of the WOCO and COLLECT

decision are the same. Fishe/Trost/Lurie (1981) and Tunali (1986) consider such a

type of a model with correlated error terms. In this case (15) has a more complicated

form:

(16) E(u1 |d1=1, d2=1, x1)

= d1d2(σu111v1|v2λ11 + σu111v2|v1 λ12) + d1(1-d2)(σu110v1|v2λ21 + σu110v2|v1λ22)

+ d2(1-d1)(σu101v1|v2λ31 + σu101v2|v1λ32) + (1-d1)(1-d2)(σu100v1|v2λ41 + σu100v2|v1 λ42)

=: d1d2(δ11 λ11 + δ12λ12) + d1(1-d2) (δ21 λ21 + δ22 λ22 ) + d2(1-d1) (δ31 λ31 + δ32 λ32)

+ (1-d1) (1-d2) (δ41 λ41 + δ42 λ42)

≡ δ1λ1 + δ2λ2,

where .|..σ are conditional covariances between u1 and v1 (v2) under the condition of

v2 (v1),

21

δjλj ≡ d1d2δ1j λ1j + d1(1-d2)δ2j λ2j + d2(1-d1)δ3j λ3j + (1-d1)(1-d2)δ4j λ4j, j = 1, 2,

(17) λij = (1 - 2d j) [ϕ(Z j)Φ[(1 - 2d3-j)Z3-j*]/P i] i=4-2dj–d3-j; j=1,2

with

Z3-j * = [Z 3-j - ρZj]/(1 - ρ²)1/2,

P i = Φ[(1 – 2dj)Zj, (1 – 2d3-j)Z3-j , (1 – 2dj)(1 – 2d3-j)ρ].

So we can determine the sign of λij. Under (15) as well as under (16) equation (9) has

to be extended by

(18) y1 = x1′β1 + d1γ1 + d2 γ2 + δ1 λ1 + δ2 λ2 + u1*,

where the meaning of λ1 and λ2 differs in both approaches. However, (18) can be

estimated consistently by OLS. Analogously, the productivity function (10) has to be

modified.

The theoretical model predicts that the effects of WOCO should differ between

covered and uncovered establishments. This should be modelled in the empirical

specification. A very crude procedure is the incorporation of an interaction term

between WOCO and COLLECT. However, usually this method cannot catch the total

impact. Also a saturated interaction model is not satisfying, because the unobserved

determinants should also vary between the different cases. Therefore, we consider

separated regressions for COLLECT=1 and COLLECT=0. In both cases the WOCO

effects are incorporated by a dummy variable.

This model with two switch regressions makes it necessary to test whether we have

to correct for a sample selection bias. The basic idea is that the employer rationally

22

decides on the membership in an employers’ association and hence on the coverage

by a collective agreement. As in the model profits are described by quasi-rents

(QRENT) and profit maximisation is assumed, the decision is based on

(19) d2 =COLLECT= 1 if (QRENT|COLLECT = 1) c > (QRENT|COLLECT = 0)

0 otherwise

2−

where c2=:CB is the cost of the membership in an employers’ association. If quasi-

rents are determined by

(20) y3k = (QRENT|COLLECT=k) = x3′β3k + u3k k=0,1

the latent index function to decide on collective bargaining is

(21) I2 = y31 - c3 - y30 = x 3 ′(β31 - β30 ) – x3c ′β3c + (u31 – u30 - u3c)

= z2′α2 – (u20 + u2c - u21) ≡ Z2 - v2

where c3 = x3c′β3c+u3c. The error term u3c is assumed uncorrelated with u31 and u30.

The employer chooses COLLECT=1 if the employer’s quasi-rent under collective

bargaining minus coverage costs is larger than the quasi-rent in case of no coverage.

The decision depends on observable (x3, x3c) and unobservable attributes (u31,u30,u3c).

The probability that I2 >0 is given by

(22) P(I2 > 0) = P(v2* < Z2*) = Φ(Z2*)

where v2*=v2/σ2, σ22=var(u31-u30-u3c), Z2*=(z2’α2)/σ2. Φ is the standard normal

distribution function because we assume that u31, u30 and u3c are normal distributed.

Then it follows analogously to equation (15)

(23) E(y31 | x3 , I2 > 0) = x3′β31 + σv2u31 (λ2|d2 =1)

(24) E(y30 | x3 , I2 < 0) = x3′β30 + σv2u30 (λ2|d2 =0).

23

As (λ2|d2=1)=-ϕ(Z2*)/Φ(Z2*)<0 and (λ2|d2=0)=ϕ(Z2*)/[1-Φ(Z2*)]>0, we should

expect that δ21=σv2u31<0 and δ20=σv2u30>0, if employers decide under the QRENT

criterion. The parameters δ21 and δ20 follow from the extended versions of (20) to

avoid a sample selection bias

(25) y3k = x3 ‘β3k + δ2k (λ2 |d2 = k) + u3k* = µ3k + δ2k (λ2 |d2 = k) + u3k*

where k=0,1. In practice, (25) has to be estimated by a two-step procedure. In the first

step the arguments of conditional λ2 variables (Z2*) will be determined via (12) and

(14) by probit estimates. In the second step (25) can be estimated by OLS where λ2 is

substituted by 2λ .

Not only the latent model of the COLLECT decision but also the WOCO decision

has to be considered in model (25) as we have argued in model (18). The mean of the

disturbance term in (25) is not zero if the WOCO decision is (partially) determined by

variables which are also relevant for QRENT. Observed and unobserved determinants

of wages are candidates. They have influence on both, QRENT - see (4) - and the

worker’s UTILITY - see (3). This means equation (25) has to be extended in the same

way as (18).

We should also determine the wage (WAGE) and the productivity (PROD)

function in this way, separately for COLLECT=1 and COLLECT=0. The arguments

for the wage function follow directly. Also, productivity correlates positively with

QRENT. Furthermore, productivity and worker’s utility are connected. Consequence

of increasing effort is higher productivity. And high effort may have two opposite

effects on worker’s utility. On the one hand, effort means usually disutility for the

workers. On the other hand, effort can be compensated by higher wages and this

effect enlarges the utility.

24

* ! $

Table 2 shows the precentage of establishments for each of the four industrial

relations regimes analysed in our theoretical model. The sample makes obvious that

all four situations are empirically relevant. Uni- and bivariate estimates of the WOCO

and COLLECT decision function are presented in Table 3. The results are used to

determine the artifical regressors λ1 and λ2 in (20). Except the coefficient for the

proportion of women and that for the proportion of skilled blue-collar in the

COLLECT function workers the uni- and bivariate estimates are althogether very

similar. As the estimated correlation coefficient between the error terms in (11) and

(12) is highly significant with ρ=0.43, we prefer the bivariate approach.

Table 4 presents the estimates of the effects of works councils and the coverage by

collective agreements on wages and productivity for all establishments. In column [1]

the WOCO effect on wages is positively significant.10 The significance holds also for

heteroscedasticity-consistent estimates (White 1980) presented in brackets. In column

[2], we consider possible correlations between unobserved determinants in (11) or

(12) and (9). Correlations between the disturbance terms in (11) and (12) are allowed.

Here, we find higher WOCO and lower COLLECT effects with larger t-statistics than

in the case of uncorrelated error terms11, while the absolute λ-effects are lower. As the

t-values exceed those under (15) and the assumption that v1 and v2 are uncorrelated

has to be rejected (see Table 3), we prefer approach (16). In both estimates the

COLLECT effect on wages is insignificant. And we find the same for the productivity

estimates – Table 4, column [3]. An exception is [4] where the COLLECT effect is

25

negatively significant. One explanation might be that collective bargaining imposes

restrictions on the establishments’ flexibility.

Tables 5 – 7 present separate estimates for establishments with and without

coverage by a collective bargaining agreement (COLLECT=1 and COLLECT=0) for

the wage (WAGE), productivity (PROD) and quasi-rent (QRENT) function.12 A

possible correlations between unobserved WOCO determinants and the disturbance

term of those functions is taken into account by λ-WOCO. The sample selection bias

induced by the COLLECT decision is corrected by λ-COLLECT.

We start with the discussion of the results in Table 5. Comparing [1] with [1a] and

[2] with [2a], the most striking result is that the impact of works councils on wages is

smaller in covered establishments than in uncovered establishments. This finding is in

accordance with the hypothesis derived from the bargaining model.

Furthermore, the model predicts that the impact of works councils on productivity

should be larger in covered establishments than in uncovered establishments. The

estimates presented in Table 6 are in accordance with this prediction. In the

benchmark model as well as in the extended model where correlated errors are taken

into account the impact of works councils on productivity is larger for the

establishments covered by collective bargaining agreements. Moreover, for the

uncovered firms the effect of works councils on productivity is insignificant.

We also note that many of the other coefficients in both specifications differ

between establishments with and without collective bargaining. The λ-WOCO and λ-

COLLECT coefficients in Table 5 and 6 are all insignificant. Nevertheless, we should

incorporate these correction variables. The comparison between column [1], [2] and

[1a], [2a], respectively, in Table 5 and 6 show that the λ variables have an impact on

the other coefficient estimates.

26

The major result of the QRENT estimates in Table 7 is that works councils have

altogether no impact on the quasi rent of an establishment, neither for firms with

collective bargaining nor for other firms. This result holds for all investigated

specifications. But the WOCO coefficients are larger under collective bargaining.

These results accord with our theoretical predictions in section III where we

concluded that the presence of a works council may result in smaller or larger profits

but that a positive influence on profitability is more likely in covered establishments.

In the latter case, the outcome δ21 <0 and δ20 >0 – see equation (27) - speaks in

favour of rational COLLECT decisions of the firms under the QRENT criterion. In

sum, all these results provide evidence that we should estimate separately for

COLLECT=1 and COLLECT=0. Our investigation shows the impact of works

councils in a new, more differentiated light than up to now. The interaction with the

bargaining situation cannot be neglected.

* $

Previous economic studies had the focus either on centralised collective bargaining or

on codetermination but not on the interaction effects between both components of the

dual industrial relations system in Germany. The theoretical analysis and the

empirical findings in this paper suggest that examining these interaction effects

contributes substantially to our understanding of the impact of industrial relations on

economic performance. There is an important indirect effect of centralised collective

bargaining. Centralised collective bargaining reduces distributional conflicts on the

establishment level. Our empirical results show that the impact of works councils on

27

wages is less strong in covered establishments compared to uncovered establishments.

Moreover, our theoretical model predicts that productivity enhancing effects of works

councils are more likely in covered establishments. We also found evidence for this

hypothesis. This evidence is in accordance with our previous findings that there is a

positive impact of councils on pay-for-performance schemes in covered but not in

uncovered establishments (Heywood, Hübler and Jirjahn 1998; Heywood and Jirjahn

1999). However, examining the interactions effects of works councils with collective

bargaining on other specific work practices such as innovations, training and the

reorganisation of work stands as important future research now.

28

+

1. Bertrand (1999) presents empirical evidence that firms that are subject to more product market

competition are more likely to renege on implicit risk sharing agreements with their employees.

2. See Levine and Tyson (1990) for a discussion of additional market failures in employee decision

making participation.

3. For the United States Klein (1984) presents empirical evidence that supervisors fear the loss of

status and power due to employee involvement programs.

4. Each wage group specifies a certain wage level. Therefore, a worker’s remuneration depends on

the assignment to a particular wage group. The assignment to a wage group usually is based on the

difficulty of the job and on the required skills.

5. The WCA provides that expenses of the works council apparatus are borne by the employer.

However, it is reasonable to assume that especially the informal costs are borne by the employees.

6. Models of bargaining over work practices and wages have also been used by Haskel (1991),

Nickell, Wadhwani and Wall (1992), and Nickell and Nicolitsas (1997). However, these models do

not consider the case that bargaining over particular work practices is only possible when there is

some form of employee representation.

7. It is well known that the disagreement point in Nash’s bargaining solution should be identified

with the impasse point of Rubinstein’s alternating-offers model (see Muthoo 1999). In this sense

αF(N) is the employer’s payoff obtained in case of perpetual disagreement.

8. Milgrom and Roberts (1992, p. 222) illustrate a similar interpretation of the worker’s personal cost

function: The incentive effects of a pay-for-performance scheme depend on the organisation of

work. For a worker working under a fixed rate production line it is difficult or impossible to

increase his own output in response to a piece-rate scheme. Consequently, workers are more

willing to co-operate with introduction of a piece-rate scheme when they have some discretion

about the pace of work.

9. Since wage negotiations between managers and councils usually are not authorised by law these

negotiations may be characterised as informal.

10. As Addison, Siebert, Wagner and Wei (2000) have restricted their estimates also on middle size

firms we have replicated this approach extended by the COLLECT effect. Quantitatively, the

WOCO effect is a little bit larger than in Table 4, [1]. The COLLECT effect is smaller and has the

expected sign in both estimates but the impact is insignificant. The estimates of the other

coefficients as well as those where a random effects estimator is used differ only slightly from

Tablee 4, [1]. The latter specification does not incorporate industries dummies because otherwise

the estimator does not converge and a comparison with fixed effects estimators is difficult. As the

industry effect cannot be neglected, in the following we focus our investigations on pooled data

with all establishments but without specific panel estimates.

11. The latter estimates are not presented in the Tables.

29

12. As the German government plans to strengthen the participatory rights of councils in smaller

establishments separate estimates for firm size classes would be interesting. However, our sample

is too small to obtain valid results.

$"

Addison, J.T., C. Schnabel and J. Wagner (1998), “Works Councils in Germany: Their Effects on Firm

Performance,” Universität Lüneburg, Arbeitsbericht Nr. 194.

Addison, J.T., C. Schnabel and J. Wagner (1997), “On the Determinants of Mandatory Works Councils

in Germany,” ! Vol. 36, pp. 419-445.

Addison, J.T., W.S. Siebert, J. Wagner and X. Wei (2000), “Worker Participation and Firm

Performance: Evidence from Germany and Britain,” "# ! Vol.

38, pp. 7-48.

Askildsen, J.E., U. Jirjahn and S.C. Smith (2000), “Works Councils and Environmental Investment:

Theory and Evidence from German Panel Data,” Working Paper, University of Hannover.

Bertrand, M. (1999), “From the Invisible Handshake to the Invisible Hand? How Import Competition

Changes the Employment Relationship,” NBER Working Paper No. 6900.

Brand, R., V. Carstensen, K. Gerlach and T. Klodt (1996), “The Hannover Panel,” Discussion Paper

No. 2, University of Hannover, Germany.

Bull, C. (1987), “The Existence of Self-Enforcing Implicit Contracts,” $

%Vol. 52, pp. 147-159.

Cooke, W. (1994), “Employee Participation Programs, Group-Based Incentives, and Company

Performance: A Union-Nonunion Comparison,” &! !'Vol. 47,

pp. 594-609.

Fishe, R.P.H., R.P. Trost and P.M. Lurie (1981), “Labor Force Earnings and College Choice of Young

Women: An Examination of Selectivity Bias and Comparative Advantage,” %

!', Vol. 1, pp. 169-191.

Fitzenberger, B. and W. Franz (1999), “Industry-Level Wage Bargaining: A Partial Rehabilitation –

The German Experience,” (# ) %, Vol. 46, pp. 437-457.

FitzRoy, F.R. and K. Kraft (2000), “Co-Determination, Efficiency and Productivity,” Working Paper,

University of Essen, Germany.

FitzRoy, F.R. and K. Kraft (1990), “Innovation, Rent-Sharing and the Organisation of Labour in the

Federal Republic of Germany,” (%" %Vol. 2, 95-103.

FitzRoy, F.R. and K. Kraft (1987), “Efficiency and Internal Organization: Works Councils in West

German Firms,” %Vol. 54, 493-504.

30

Freeman, R.B. and R. Gibbons (1995), “Getting Together and Breaking Apart: The Decline of

Centralized Collective Bargaining,” in * # + +( , edited by R.B.

Freeman and L.F. Katz, Chicago: University of Chicago Press, pp. 345-370.

Freeman, R.B. and E.P. Lazear (1995), “An Economic Analysis of Works Councils,” in ,

- !. . ! edited by J.

Rogers and W. Streeck, Chicago: University of Chicago Press, pp. 27-52.

Freeman, R.B. and J.L. Medoff (1984), #*/ *0New York: Basic Books.

Frick, B. and D. Sadowski (1995), “Works Councils, Unions and Firm Performance: The Impact of

Workers’ Participation in Germany,” in 1%, &2,)%

%.' 3 # /4(4 5% % edited by F. Buttler, W. Franz, R.

Schettkat, and D. Soskice, New York: Routledge, pp. 46-81.

Haskel, J. (1991), “Imperfect Competition, Work Practices and Productivity Growth,” 6"

% (Vol. 53, pp. 265-279.

Heywood, J.S., O. Hübler and U. Jirjahn (1998), “Variable Payment Schemes and Industrial Relations:

Evidence from Germany,” ,Vol. 51, pp. 237-257.

Heywood, J.S. and U. Jirjahn (1999), “Payment Schemes, Gender and Industrial Relations in

Germany,” Working Paper, University of Hannover.

Hübler, O. and W. Meyer (2000), “Industrial Relations and the Wage Differentials between Skilled and

Unskilled Blue-Collar Workers within Establishments: An Empirical Analysis with Data of

Manufacturing Firms,” IZA Discussion Paper No. 176.

Jirjahn, U. and T. Klodt (1999), “Lohnhöhe, industrielle Beziehungen und Produktmärkte,” in 7

, + ' # "#8+ + " ' "&9 / #%

edited by L. Bellmann, S. Kohaut and M. Lahner, Nürnberg: BeitrAB, pp. 27-54.

Klein, J. (1984), “Why Supervisors Resist Employee Involvement,” :'" !', pp. 87-

95.

Kreps, D.M. (1990), “Corporate Culture and Economic Theory,” in ).' )')

%edited by J.E. Alt and K.A. Shepsle, Cambridge: Cambridge University Press, pp.90-143.

Levine, D.I. and L. Tyson (1990), “Participation, Productivity, and the Firm’s Environment,” in ) +

) 'edited by A.S. Blinder, Washington, D.C.: Brookings Institution, pp. 183-237.

Lindbeck, A. and D.J. Snower (1997), “Centralised Bargaining, Multi-Tasking and Work Incentives,”

Centre for Economic Policy Research, Discussion Paper No. 1563.

McCain, R.A. (1980), “A Theory of Codetermination,” 7#;< =, %Vol. 40, pp.

65-90.

Milgrom, P. and J. Roberts (1992), %+ > 2 +% New Jersey: Prentice-

Hall International.

Moene, K.O., M. Wallerstein and M. Hoel (1993), “Bargaining Structure and Economic Performance,”

in / "#' )9"+ + %)% edited by R.J. Flanagan,

K.O. Moene and M. Wallerstein, Oxford: Clarendon Press, pp. 63-131.

31

Müller-Jentsch, W. (1995), “Germany: From Collective Voice to Co-Management,” in ,

- !. . ! edited by J. Rogers and W.

Streeck, Chicago: University of Chicago Press, pp. 53-78.

Muthoo, A. (1999), "+ +##?.. , Cambridge: Cambridge University Press.

Nickell, S. and D. Nicolitsas (1997), “Wages, Restrictive Practices and Productivity,” &

%Vol. 4, pp. 201-221.

Nickell, S., S. Wadhwani and M. Wall (1992), “Productivity Growth in U.K. Companies, 1975-1986,”

. %!'Vol. 36, pp.1055-1091.

Prendergast, C. (1999), “The Provision of Incentives in Firms,” % Vol.

37, pp. 7-63.

Smith, S.C. (1991), “On the Economic Rationale for Codetermination Law,” %

"#' + > Vol. 12, pp. 261-281.

Svejnar, J. (1982), “On the Theory of a Participatory Firm,” %#Vol. 27, pp.

313-330.

Tunali, I. (1986), “A General Structure for Models of Double-Selection and an Application to a Joint

Migration/Earnings Process with Remigration,” in !# & %Vol. 8 (Part B),

edited by R.G. Ehrenberg, JAI Press, pp. 235-283.

White, H. (1980), “A Heteroscedasticity-Consistent Covariance Matrix Estimator and a Direct Test for

Heteroscedasticity,” %, Vol. 48, pp.817-83.

32

Em

ploy

er

Wor

kers

Wor

kers

No

cove

rage

(C

OL

LE

CT

= 0

) C

over

age

(CO

LL

EC

T =

1)

No

wor

ks c

ounc

il (W

OC

O =

0)

Wor

ks c

ounc

il (W

OC

O =

1)

No

wor

ks c

ounc

il (W

OC

O =

0)

Wor

ks c

ounc

il (W

OC

O =

1)

•

• •

π(e*

,w*)

= F

(N),

N·U

(e*,

w*)

= 0

π(e*

,w*)

– C

B =

F(N

) –

CB,

N·U

(e*,

w*)

= 0

π(e*

**,w

***)

– C

B =

(1

+F(

N)/

3θ

α+

2~

) –

CB,

N·U

(e**

*,w

***)

–C

W =

(1

+F(

N)/

9θ

α−

~)2

– C

W

π(e*

*,w

**)

= (

1+

F(N

)/3

θ),

N·U

(e**

,w**

) –

CW

= (

1+

F(N

)/9

θ)2

– C

W

33

0 <θ < 2 (1 )⋅ − ~α

2 (1 )⋅ − ~α < θ < 2

2 < θ

CW

CB 0

COLLECT = 0, WOCO = 0

COLLECT = 1, WOCO = 0

COLLECT = 1, WOCO = 0

COLLECT = 0, WOCO = 1

COLLECT = 1, WOCO = 1

COLLECT = 0, WOCO = 1

COLLECT = 0, WOCO = 1

2 F(N) / 3~α (2 )F(N) / 3− θ

(1+

9F(N)

θ α− ~)2

(1+ )F(N)

2θ9

CW

CB 0

COLLECT = 0, WOCO = 0

COLLECT = 1, WOCO = 0

COLLECT = 0, WOCO = 0

COLLECT = 0, WOCO = 1

COLLECT = 1, WOCO = 1

COLLECT = 1, WOCO = 1

COLLECT = 0, WOCO = 1

2 F(N) / 3~α (2 )F(N) / 3− θ

(1+

9F(N)

θ α− ~)2

(1+ )F(N)

2θ9

CW

CB 0

COLLECT = 0, WOCO = 0

COLLECT = 0, WOCO = 1

COLLECT = 1, WOCO = 1

COLLECT = 0, WOCO = 1

2 F(N) / 3~α

(1+

9F(N)

θ α− ~)2

(1+ )F(N)

2θ9

34

Subgame-perfect equilibrium, wage rate, productivity and profitability per head

Strategy pair Conditions for a subgame-

perfect equilibrium Wage rate

W

Productivity

(1+e)F(N)/N

Profitability per head

[(1+e)F(N)–wN]/N

COLLECT=0,

WOCO =0

CB > 0,

CW > (1+ ) F(N) / 92θ

0

F(N)/N

F(N)/N

COLLECT=0,

WOCO = 1

CB > 2 F(N)~α /3,

CW < (1 + ) F(N) / 92θ α− ~

or

CB > (2 )F(N) / 3− θ ,

(1+ ) F(N) / 92θ > CW >

(1 + ) F(N) / 92θ α− ~

(1 + )F(N)

3Nθ

(1 + )2F(N)

3Nθ

(1 + )F(N)

3Nθ

COLLECT=1, WOCO=0

CB < (2 )F(N) / 3− θ ,

(1+ ) F(N) / 92θ > CW >

(1 + ) F(N) / 92θ α− ~

0

F(N)/N

F(N)/N

COLLECT=1,

WOCO = 1

CB < 2 F(N)~α ,

CW < (1+ ) F(N) / 92θ α− ~

(1 + )F(N)

3Nθ α− ~

(1 + )2F(N)

3Nθ α+ 0 5. ~

(1 + )F(N)

3Nθ α+ 2~

: Empirical distribution of establishments across industrial relations regimes (Percent of establishments) WOCO = 1

47.6 10.6

WOCO = 0 18.4 23.4

35

: Probit ML estimates of works councils and coverage by collective bargaining agreement,

absolute t statistics in parentheses

Constant -1.7436(7.12) -1.7242(6.15) 0.1406(0.65) 0.1034(0.47)

Firm size (number of employees) 0.0104(12.96) 0.0104(21.32) 0.0031(6.74) 0.0032(6.85)

Firm created before 1960 0.6164(6.08) 0.5902(5.35) 0.6504(7.49) 0.6634(7.40)

Single firm (no subsidiaries) -0.6344(4.87) -0.6041(4.14) -0.2595(2.77) -0.2511(2.62)

Wages above industry agreement 0.2494(2.41) 0.2460(2.17) 0.3619(4.05) 0.3648(3.93)

Use of piece rates 0.6901(4.59) 0.6858(4.30) 0.2925(2.40) 0.2932(2.42)

Proportion of women -0.8745(2.98) -0.8265(2.45) -0.6386(2.68) -0.6610(2.49)

Flexible working time arrangements 0.2188(2.20) 0.2240(2.14)

Use of newest production technology -0.3726(3.63) -0.3701(3.34) -0.2200(2.44) -0.2170(2.29)

Team work for blue collar workers -0.2781(2.92) -0.2589(2.53)

Proportion of skilled blue collar workers 0.3838(2.06) 0.4543(2.47)

Employer provided further training 0.3424(3.42) 0.3307(3.22)

Proportion of part-time workers -1.6694(3.34) -1.6342(3.10)

15 Industry dummies + + + +

ρ 0.4351(7.75)

χ² 734 261

Log-likelihood -585 -1051 -614

N 1276 1258 1276 1258

: Wage and productivity estimates with respect to works councils and coverage by collective bargaining agreement effects in all establishments, absolute t statistics in parentheses, heteroscedastic-consistent estimates in brackets

Works council (WOCO) 4.1439

(3.36) [2.10]

4.3961

(2.63) [1.60]

15.6197

(2.65) [2.50]

23.0383

(2.61) [2.68]

Coverage by a collective agreement

(COLLECT)

1.3108

(1.15) [1.00]

1.3808

(0.63) [0.72]

-8.4982

(1.45) [1.42]

-35.2500

(3.06) [2.81]

λ-WOCO -0.2276

(0.25) [0.27]

-4.6825

(0.84) [1.00]

λ-COLLECT -0.4797

(0.42) [0.45]

10.0413

(1.69) [1.90]

R² 0.288 0.288 0.255 0.283

N 910 910 816 732

Control variables: All variables listed in Table 5 are included in the wage regression. All variables listed in Table 6 are included in the productivity regression.

36

: Wage estimates with respect to works councils in firms with and without coverage by collective bargaining agreements, absolute t statistic in parentheses, heteroscedastic-consistent estimates in brackets

Constant

48.9933

(5.55) [6.07]

48.8906

(5.37) [6.01]

44.2142

(3.95) [4.08]

36.6803

(2.42) [2.51]

Firm size (number of employees)

0.0021

(1.19) [2.00]

0.0021

(0.76) [1.14]

-0.0071

(0.49) [0.63]

-0.0122

(0.63) [0.88]

Single firm (no subsidiaries) -3.3746

(1.86) [2.26]

-3.1644

(1.63) [1.97]

-2.7209

(1.31) [1.42]

-1.6615

(0.54) [0.67]

Use of newest production technology 1.5604

(1.15) [1.22]

1.2800

(0.92) [0.91]

2.1775

(1.29) [1.36]

3.4325

(1.52) [1.64]

Proportion of women -19.7172

(5.03) [6.81]

-19.7496

(4.86) [6.31]

-21.9068

(4.69) [4.94]

-20.4599

(3.12) [3.52]

Team work for blue collar workers -1.1661

(0.92) [0.87]

-1.0172

(0.77) [0.76]

-1.9409

(1.22) [1.32]

-2.2340

(1.09) [1.23]

Tenure of the work force

0.0745

(2.66) [2.09]

0.0755

(2.64) [1.99]

-0.0326

(0.93) [0.97]

-0.0987

(2.13) [2.20]

Proportion of part-time workers

-14.1879

(1.81) [2.43]

-13.4781

(1.61) [2.18]

-20.7583

(3.10) [2.86]

-15.4382

(1.71) [1.62]

Overtime hours 0.5917

(1.76) [1.81]

0.5610

(1.62) [1.66]

0.7711

(2.32) [2.59]

0.5246

(1.05) [1.28]

Shiftwork 6.2139

(4.17) [3.15]

5.8048

(3.56) [2.46]

2.3687

(1.18) [1.39]

0.2984

(0.11) [0.15]

Proportion of university graduates 12.5120

(0.81) [1.00]

16.2128

(1.00) [1.31]

30.0091

(1.97) [1.61]

-4.6444

(0.23) [0.26]

Time dummy 1996

1.7745

(1.43) [1.42]

1.6708

(1.31) [1.26]

-0.4791

(0.31) [0.33]

0.9655

(0.46) [0.51]

Works council (WOCO)

3.9894

(2.39) [1.31]

5.8530

(2.00) [1.47]

5.8970

(2.86) [2.83]

7.2398

(2.52) [2.84]

λ-WOCO -1.6627

(0.90) [1.04]

0.3266

(0.23) [0.27]

λ-COLLECT

1.3895

(0.60) [0.50]

0.0937

(0.05) [0.06]

15 industry dummies + + + +

R² 0.325 0.324 0.321 0.301

N 682 663 359 247

37

: Productivity estimates with respect to works councils in firms with and without coverage by collective bargaining agreements, absolute t statistics in parentheses, heteroscedastic-consistent estimates in brackets

Constant 168.0142

(7.56) [7.17]

163.4020

(7.44) [6.74]

223.5868

(6.65) [6.23]

239.5144

(4.76) [4.76]

Firm size (number of employees)

0.0045

(0.97) [1.74]

0.0024

(0.55) [1.27]

0.1508

(2.21) [1.19]

0.2070

(2.38) [1.16]

Firm created before 1960

-10.4713

(1.48) [1.09]

-2.9556

(0.37) [0.28]

-19.3712

(2.01) [2.16]

-9.3392

(0.73) [0.85]

Single firm (no subsidiaries)

-4.7430

(0.76) [0.71]

-7.3937

(1.17) [1.05]

-16.0460

(1.49) [1.56]

-19.9792

(1.57) [1.59]

Limited company (legal form) 10.3839

(1.60) [1.48]

10.4852

(1.69) [1.52]

9.4716

(0.88) [0.99]

17.5947

(1.38) [1.36]

Use of newest production technology 23.7848

(3.78) [3.71]

21.3918

(3.53) [3.54]

28.5686

(2.94) [3.27]

32.8590

(2.79) [2.78]

Market share of the firm 0.5005

(3.14) [2.83]

0.4756

(3.14) [2.71]

0.0061

(0.02) [0.03]

-0.4452

(1.41) [1.56]

Proportion of women -67.5055

(3.67) [3.91]

-67.6881

(3.81) [4.18]

-6.0548

(0.21) [0.21]

-34.6184

(1.02) [0.95]

Team work for blue collar workers -8.1836

(1.31) [1.39]

-6.6949

(1.12) [1.18]

-21.3465

(2.33) [2.45]

-11.2683

(1.05) [1.24]

Proportion of part-time workers -34.2368

(0.90) [1.06]

-30.3267

(0.83) [0.95]

-116.5645

(2.75) [3.01]

-81.2745

(1.76) [2.04]

Proportion of university graduates 105.9964

(1.47) [1.66]

88.7520

(1.29) [1.39]

115.2669

(1.46) [1.34]

159.8195

(1.48) [1.55]

Proportion of blue collar workers -12.5543

(0.55) [0.53]

-26.8247

(1.24) [1.21]

-65.5652

(2.29) [2.23]

-67.4491

(1.82) [1.88]

Proportion of apprentices -231.4640

(3.76) [4.72]

-240.7144

(4.13) [4.72]

-192.6515

(2.53) [3.43]

-186.7010

(2.02) [2.55]

Profit sharing for employees -2.1179

(0.26) [0.30]

2.8917

(0.37) [0.43]

33.4227

(2.84) [3.04]

33.6692

(2.57) [3.04]

Time dummy 1996

13.2468

(2.21) [2.27]

11.2035

(1.97) [2.02]

1.0359

(0.11) [0.12]

5.5428

(0.49) [0.55]

Works council (WOCO)

20.3769

(2.55) [2.51]

27.6394

(2.49) [2.66]

12.3770

(1.03) [0.87]

5.8622

(0.29) [0.26]

λ-WOCO -8.3151

(1.17) [1.52]

0.0787

(0.01) [0.01]

λ-COLLECT 3.7280

(0.31) [0.44]

10.3473

(0.86) [1.02]

15 industry dummies + + + +

R² 0.258 0.273 0.385 0.486

N 566 555 258 177

38

: Quasi-rent estimates with respect to works councils in firms with and without coverage by collective bargaining agreement, absolute t statistics in parentheses, heteroscedastic-consistent estimates in brackets

Works council (WOCO)

10.2706

(1.28) [1.26]

15.7507

(1.40) [1.51]

4.5160

(0.39) [0.31]

-0.8770

(0.04) [0.04]

λ-WOCO -6.68

(0.94) [1.28]

-4.2923

(0.36) [0.41]

λ-COLLECT 3.7495

(0.31) [0.44]

7.0068

(0.60) [0.70]

R² 0.221 0.228 0.386 0.518

N 503 490 211 130

Control variables: All variables listed in Table 6 are included.

241

W. Koeniger

Trade, Labor Market Rigidities, and Government-Financed Technological Change

2 01/01

242

G. Faggio J. Konings

Job Creation, Job Destruction and Employment Growth in Transition Countries in the 90’s

4 01/01

243

E. Brainerd

Economic Reform and Mortality in the Former Soviet Union: A Study of the Suicide Epidemic in the 1990s

4 01/01

244

S. M. Fuess, Jr. M. Millea

Pay and Productivity in a Corporatist Economy: Evidence from Austria

5 01/01

245

F. Andersson K. A. Konrad

Globalization and Human Capital Formation

5 01/01

246

E. Plug W. Vijverberg

Schooling, Family Background, and Adoption: Does Family Income Matter?

5 01/01

247

E. Plug W. Vijverberg

Schooling, Family Background, and Adoption: Is it Nature or is it Nurture?

5 01/01

248

P. M. Picard E. Toulemonde

The Impact of Labor Markets on Emergence and Persistence of Regional Asymmetries

2 01/01

249

B. M. S. van Praag P. Cardoso

“Should I Pay for You or for Myself?” The Optimal Level and Composition of Retirement Benefit Systems

3 01/01

250

T. J. Hatton J. G. Williamson

Demographic and Economic Pressure on Emigration out of Africa

1 01/01

251

R. Yemtsov

Labor Markets, Inequality and Poverty in Georgia

4 01/01

252

R. Yemtsov

Inequality and Income Distribution in Georgia

4 01/01

253

R. Yemtsov

Living Standards and Economic Vulnerability in Turkey between 1987 and 1994

4 01/01

254

H. Gersbach A. Schniewind

Learning of General Equilibrium Effects and the Unemployment Trap

3 02/01

255

H. Gersbach A. Schniewind

Product Market Reforms and Unemployment in Europe

3 02/01

256

T. Boeri H. Brücker

Eastern Enlargement and EU-Labour Markets: Perceptions, Challenges and Opportunities

2 02/01

257

T. Boeri

Transition with Labour Supply

4 02/01

258

M. Rosholm K. Scott L. Husted

The Times They Are A-Changin’: Organizational Change and Immigrant Employment Opportunities in Scandinavia

1 02/01

259

A. Ferrer-i-Carbonell B. M.S. van Praag

Poverty in the Russian Federation

4 02/01

260

P. Cahuc F. Postel-Vinay

Temporary Jobs, Employment Protection and Labor Market Performance

1/3 02/01

261

M. Lindahl

Home versus School Learning: A New Approach to Estimating the Effect of Class Size on Achievement

5 02/01

262

M. Lindahl

Summer Learning and the Effect of Schooling: Evidence from Sweden

5 02/01

263 N. Datta Gupta N. Smith

Children and Career Interruptions: The Family Gap in Denmark

5 02/01

264 C. Dustmann Return Migration, Wage Differentials, and the

Optimal Migration Duration 1 02/01

265 M. Rosholm

M. Svarer

Structurally Dependent Competing Risks

1 02/01

266 C. Dustmann O. Kirchkamp

The Optimal Migration Duration and Activity Choice after Re-migration

1 02/01

267 A. Newell

The Distribution of Wages in Transition Countries

4 03/01

268 A. Newell B. Reilly

The Gender Pay Gap in the Transition from Communism: Some Empirical Evidence

4 03/01

269 H. Buddelmeyer

Re-employment Dynamics of Disabled Workers

3 03/01

270 B. Augurzky

C. M. Schmidt

The Evaluation of Community-Based Interventions: A Monte Carlo Study

6 03/01

271 B. Augurzky

C. M. Schmidt

The Propensity Score: A Means to An End 6 03/01

272 C. Belzil J. Hansen

Heterogeneous Returns to Human Capital and Dynamic Self-Selection

5 03/01

273 G. Saint-Paul

Distribution and Growth in an Economy with Limited Needs

5 03/01

274 P. J. Pedersen N. Smith

Unemployment Traps: Do Financial Dis-incentives Matter?

3 03/01

275 G. S. Epstein T. Lecker

Multi-Generation Model of Immigrant Earnings: Theory and Application

1 03/01

276 B. Amable D. Gatti

The Impact of Product Market Competition on Employment and Wages

5 03/01

277 R. Winter-Ebmer Evaluating an Innovative Redundancy-Retraining

Project: The Austrian Steel Foundation

6 03/01

278 T. M. Andersen

Welfare Policies, Labour Taxation and Inter-national Integration

2 04/01

279 T. M. Andersen

Product Market Integration, Wage Dispersion and Unemployment

2 04/01

280 P. Apps R. Rees

Household Saving and Full Consumption over the Life Cycle

7 04/01

281 G. Saint-Paul

Information Technology and the Knowledge Elites

5 04/01

282 J. Albrecht A. Björklund S. Vroman

Is There a Glass Ceiling in Sweden?

5 04/01

283 M. Hagedorn A. Kaul V. Reinthaler

Welfare Analysis in a Schumpeterian Growth Model with Capital

7 04/01

284 H. Rapoport A. Weiss

The Optimal Size for a Minority

1 04/01

285 J. Jerger C. Pohnke A. Spermann

Gut betreut in den Arbeitsmarkt? Eine mikroökonometrische Evaluation der Mannheimer Arbeitsvermittlungsagentur

5 04/01

286 M. Fertig C. M. Schmidt

First- and Second-Generation Migrants in Germany –What Do We Know and What Do People Think

1 04/01

287 P. Guggenberger A. Kaul M. Kolmar

Efficiency Properties of Labor Taxation in a Spatial Model of Restricted Labor Mobility

3 04/01

288 D. A. Cobb-Clark

Getting Ahead: The Determinants of and Payoffs to Internal Promotion for Young U.S. Men and Women

5 04/01

289 L. Cameron D. A. Cobb-Clark

Old-Age Support in Developing Countries: Labor Supply, Intergenerational Transfers and Living Arrangements

3 04/01

290

D. A. Cobb-Clark M. D. Connolly C. Worswick

The Job Search and Education Investments of Immigrant Families

1 04/01

291

R T. Riphahn

Cohort Effects in the Educational Attainment of Second Generation Immigrants in Germany: An Analysis of Census Data

1 05/01

292

E. Wasmer

Between-group Competition in the Labor Market and the Rising Returns to Skill: US and France 1964-2000

5 05/01

293

D. Cobb-Clark T. F. Crossley

Gender, Comparative Advantage and Labor Market Activity in Immigrant Families

1 05/01

294

Š. Jurajda

Estimating the Effect of Unemployment Insurance Compensation on the Labor Market Histories of Displaced Workers

3 05/01

295

F. Duffy P. P. Walsh

Individual Pay and Outside Options: Evidence from the Polish Labour Force Survey

4 05/01

296

H. S. Nielsen M. Rosholm N. Smith L. Husted

Intergenerational Transmissions and the School-to-Work transition of 2nd Generation Immigrants

1 05/01

297

J. C. van Ours J. Veenman

The Educational Attainment of Second Generation Immigrants in The Netherlands

1 05/01

298

P. Telhado Pereira P. Silva Martins

Returns to Education and Wage Equations 5 06/01

299

G. Brunello C. Lucifora R. Winter-Ebmer

The Wage Expectations of European College Students

5 06/01

300

A. Stutzer R. Lalive

The Role of Social Work Norms in Job Searching and Subjective Well-Being

5 06/01

301

J. R. Frick G.G. Wagner

Economic and Social Perspectives of Immigrant Children in Germany

1 06/01

302

G. S. Epstein A. Weiss

A Theory of Immigration Amnesties

1 06/01

303

G. A. Pfann B. F. Blumberg

Social Capital and the Uncertainty Reduction of Self-Employment

5 06/01

304

P. Cahuc E. Wasmer

Labour Market Efficiency, Wages and Employ-ment when Search Frictions Interact with Intra-firm Bargaining

2 06/01

305

H. Bonin

Fiskalische Effekte der Zuwanderung nach Deutschland: Eine Generationenbilanz

1 06/01

306

H. Bonin G. Abío E. Berenguer J. Gil C. Patxot

Is the Deficit under Control? A Generational Accounting Perspective on Fiscal Policy and Labour Market Trends in Spain

2 06/01

307

G. A. Pfann

Downsizing

1/5 06/01

308

G. A. Pfann D. S. Hamermesh

Two-Sided Learning, Labor Turnover and Worker Displacement

1 06/01

309

G. Brunello On the Complementarity between Education and Training in Europe

5 06/01

310

U. Sunde Human Capital Accumulation, Education and Earnings Inequality

5 06/01

311

G. Brunello Unemployment, Education and Earnings Growth

3 06/01

312

C. Furnée M. Kemler G. A. Pfann

The Value of Pain Relief

5 06/01

313

A. Ferrer-i-Carbonell B. M.S. van Praag

The Subjective Costs of Health Losses due to Chronic Diseases: An Alternative Model for Monetary Appraisal

7 06/01

314

B. M.S. van Praag A. Ferrer-i-Carbonell

Age-Differentiated QALY Losses

7 06/01

315

W. H. J. Hassink R. Schettkat

On Price-Setting for Identical Products in Markets without Formal Trade Barriers

7 06/01

316

M. Frondel C. M. Schmidt

Rejecting Capital-Skill Complementarity at all Costs

5 06/01

317

R. Winkelmann

Health Care Reform and the Number of Doctor Visits – An Econometric Analysis

7 06/01

318

M. Pannenberg G. G. Wagner

Overtime Work, Overtime Compensation and the Distribution of Economic Well-Being: Evidence for West Germany and Great Britain

1 06/01

319

R. Euwals R. Winkelmann

Why do Firms Train? Empirical Evidence on the First Labour Market Outcomes of Graduated Apprentices

1 06/01

320

R. Fahr U. Sunde

Strategic Hiring Behavior in Empirical Matching Functions

1 06/01

321

P. Telhado PereiraP. Silva Martins

Is there a Return – Risk Link in Education?

5 07/01

322

O. Hübler U. Jirjahn

Works Councils and Collective Bargaining in Germany: The Impact on Productivity and Wages

1 07/01

An updated list of IZA Discussion Papers is available on the center‘s homepage .